ChatGPT may introduce a separate mode to protect stored conversations from prying eyes. Such chats can be hidden from history and searches, and can only be accessed after entering a PIN or verifying a fingerprint.
Image source: OpenAI (modified)
Reporters found that the ChatGPT application of Android version 1.2026.230 has signs of developing new features Android Authority. The text strings in the code describe in detail how Secure Chat operates. Specifically, once blocked, conversation titles and content will no longer appear in ChatGPT’s regular sidebar, history, and search results. To open the conversation again, you will need a PIN or fingerprint. In this case, successful authorization will not cancel the protection and the chat will remain blocked until the user switches it back to normal mode themselves.
As Android Authority points out, this innovation should not be confused with the ad-hoc chats that are already available. The latter are not saved in the history, so you cannot return to them after the conversation ends, while blocked letters will be stored in the account and will still be available to the owner. Judging from the description, any saved chat history can be protected in this way.
ChatGPT also provides a special way to save data. Locked chats will be able to take into account information in shared memory, but new information saved by the neural network during secure communication will not be used in regular conversations and will remain in a separate protected chat. Therefore, information from such chats does not automatically enter the ChatGPT shared memory.
The new version already contains a fairly detailed description of the upcoming feature, but OpenAI hasn’t announced the rollout yet, so it’s unclear if or when it will be available to users.
If you find an error, select it with your mouse and press CTRL+ENTER.